For example: I can set a room clear script on a boss fight room that sets BOSS_DEFEATED_0 (or whatever), and make the boss's spawn conditional on that flag being clear, so bosses don't respawn when you revisit a room.

Consolation prize: My micro optimizations freed enough memory to make entity spawns conditional on global flags, which makes a lot of things a lot simpler to implement without using more memory.

Ooof. This is generating more bugs than I can fix. Fuck it. Overlay experiment over, I'm pulling the plug. I will just have to accept that there are some limits I can't break.

Moving play-mode-specific functions into the overlay segment is how we actually claw back the memory spent on the overlay slot itself, freeing that much memory in the program segment for resident code.

My next attempt is to gather up some play mode functions and move them into a play-mode-specific overlay, which should clear out some space.

So far, I haven't actually saved any significant memory in the program segment with overlays. What space has been freed has been fed to the overlay segment. I'll keep at it a while, but if I don't see a real return on this, I'll be pissed at having to revert it.

I swear to Odin, when the full and final version of this game is released, I'm going to burn every copy of the source code. This shit is cursed.

In Turbo Pascal 3, you could just mark functions "overlay" and the compiler handled it all. No such joy here.

The real savings will kick in when I find the right size for the code overlay segment, so that sufficient play mode rendering code can live in the overlay (as it's mutually exclusive with other modes!) thus actually freeing that space for feature work and bug fixes.

Might as well be productive if I can't sleep. Find something else to stick in a code overlay, maybe free up some space.

Yay. Anxiety induced insomnia. Fuck my life.

This overlay bollocks had better pay off.